What is claimed is:1. A spectral characteristics prediction method for predicting spectral characteristics obtained by applying ink on a base material, the spectral characteristics prediction method comprising:a prediction target color setting step of setting, to a prediction target color, an ink color for which spectral characteristics of a maximum gradation value, spectral characteristics of a minimum gradation value, and spectral characteristics of at least one intermediate gradation value are obtained;a first relational equation calculation step of obtaining a first relational equation that, with an intermediate gradation value for which spectral characteristics are obtained regarding the prediction target color being taken as a characteristics-acquired gradation value, represents a relationship between spectral characteristics of the maximum gradation value and spectral characteristics of the characteristics-acquired gradation value for each of a plurality of sample colors which are a plurality of ink colors each of which spectral characteristics of the maximum gradation value, spectral characteristics of the minimum gradation value, and spectral characteristics of at least one intermediate gradation value are obtained;a first prediction step of, for each of the plurality of sample colors, obtaining prediction values of spectral characteristics of the characteristics-acquired gradation value for the prediction target color by applying the spectral characteristics of the maximum gradation value for the prediction target color to a corresponding first relational equation;a difference value calculation step of, for each of the plurality of sample colors, obtaining a difference value between the prediction values obtained in the first prediction step and actual measurement values of spectral characteristics of the characteristics-acquired gradation value for the prediction target color;a reference color selection step of selecting, as a reference color, a sample color for which a minimum difference value is obtained in the difference value calculation step among the plurality of sample colors;a second relational equation calculation step of, with the maximum gradation value or the characteristic-acquired gradation value being taken as a reference gradation value, and with a gradation value for which spectral characteristics are obtained regarding the reference color or the characteristics-acquired gradation value being taken as a process target gradation value, obtaining a second relational equation that represents a relationship between spectral characteristics of the reference gradation value and spectral characteristics of the process target gradation value for the reference color; anda second prediction step of, using the second relational equation, obtaining prediction values of spectral characteristics of a prediction target gradation value for the prediction target color.2. The spectral characteristics prediction method according to claim 1, whereina number of the characteristics-acquired gradation values is one, andin the second relational equation calculation step,regarding the process target gradation value between the maximum gradation value and the characteristics-acquired gradation value, with the spectral characteristics of the maximum gradation value being taken as a first reference and with the spectral characteristics of the characteristic-acquired gradation value being taken as a second reference, an equation that represents a relationship between the spectral characteristics of the maximum gradation value and the spectral characteristics of the process target gradation value is obtained as the second relational equation, andregarding the process target gradation value between the characteristics-acquired gradation value and the minimum gradation value, with the spectral characteristics of the characteristics-acquired gradation value being taken as a first reference and with the spectral characteristics of the minimum gradation value being taken as a second reference, an equation that represents a relationship between the spectral characteristics of the characteristics-acquired gradation value and the spectral characteristics of the process target gradation value is obtained as the second relational equation.3. The spectral characteristics prediction method according to claim 2, wherein, in the second relational equation calculation step, data of the spectral characteristics are subjected to normalization so that a value of the spectral characteristics being taken as the second reference is 1.4. The spectral characteristics prediction method according to claim 1, whereinthe at least one intermediate gradation value for which the spectral characteristics are obtained regarding the prediction target color includes a first gradation value and a second gradation value smaller than the first gradation value, andin the second relational equation calculation step,regarding the process target gradation value between the maximum gradation value and the first gradation value, with the spectral characteristics of the maximum gradation value being taken as a first reference and with the spectral characteristics of the first gradation value being taken as a second reference, an equation that represents a relationship between the spectral characteristics of the maximum gradation value and the spectral characteristics of the process target gradation value is obtained as the second relational equation,regarding the process target gradation value between the first gradation value and the second gradation value, with the spectral characteristics of the first gradation value being taken as a first reference and with the spectral characteristics of the second gradation value being taken as a second reference, an equation that represents a relationship between the spectral characteristics of the first gradation value and the spectral characteristics of the process target gradation value is obtained as the second relational equation, andregarding the process target gradation value between the second gradation value and the minimum gradation value, with the spectral characteristics of the second gradation value being taken as a first reference and with the spectral characteristics of the minimum gradation value being taken as a second reference, an equation that represents a relationship between the spectral characteristics of the second gradation value and the spectral characteristics of the process target gradation value is obtained as the second relational equation.5. The spectral characteristics prediction method according to claim 1, whereinthe at least one intermediate gradation value for which the spectral characteristics are obtained regarding the prediction target color includes m pieces of gradation values from a first to as m-th gradation value with m being taken as an integer of 3 or more,a k-th gradation value is larger than a (k+1)-th gradation value with k being taken as an integer of 1 or more and (m?1) or less, andin the second relational equation calculation step,regarding the process target gradation value between the maximum gradation value and the first gradation value, with the spectral characteristics of the maximum gradation value being taken as a first reference and with the spectral characteristics of the first gradation value being taken as a second reference, an equation that represents a relationship between the spectral characteristics of the maximum gradation value and the spectral characteristics of the process target gradation value is obtained as the second relational equation,regarding the process target gradation value between the k-th gradation value and the (k+1)-th gradation value, with the spectral characteristics of the k-th gradation value being taken as a first reference and with the spectral characteristics of the (k+1)-th gradation value being taken as a second reference, an equation that represents a relationship between the spectral characteristics of the k-th gradation value and the spectral characteristics of the process target gradation value is obtained as the second relational equation, andregarding the process target gradation value between the m-th gradation value and the minimum gradation value, with the spectral characteristics of the m-th gradation value being taken as a first reference and with the spectral characteristics of the minimum gradation value being taken as a second reference, an equation that represents a relationship between the spectral characteristics of the m-th gradation value and the spectral characteristics of the process target gradation value is obtained as the second relational equation.6. The spectral characteristics prediction method according to claim 1, wherein,in the second relational equation calculation step, an equation that represents a relationship between the spectral characteristics of the maximum gradation value and the spectral characteristics of the process target gradation value for the reference color is obtained as the second relational equation, andin the second prediction step, the prediction values of the spectral characteristics of the prediction target gradation value for the prediction target color is obtained by applying the spectral characteristics of the maximum gradation value for the prediction target color to the second relational equation.7. The spectral characteristics prediction method according to claim 1, whereina number of the prediction target gradation values is two or more, andthe second prediction step includes:a second relational equation using step of obtaining prediction values of spectral characteristics of the gradation value for which the spectral characteristics are obtained regarding the reference color among two or more of the prediction target gradation values by applying spectral characteristics of the reference gradation value for the prediction target color to a corresponding second relational equation; andan interpolation step of obtaining prediction values of spectral characteristics of a gradation value for which the spectral characteristics are not obtained regarding the reference color among two or more of the prediction target gradation values by performing spline interpolation based on the prediction values obtained in the second relational equation using step.8. The spectral characteristics prediction method according to claim 1, further comprising a union creation step of creating a union of the gradation value for which spectral characteristics are obtained regarding the reference color and the gradation value for which spectral characteristics are obtained regarding the prediction target color, wherein,in the second relational equation calculation step, the gradation value included in the union is taken as the process target gradation value,a number of the prediction target gradation values is two or more, andthe second prediction step includes:a second relational equation using step of obtaining prediction values of spectral characteristics of the gradation value included in the union among two or more of the prediction target gradation values by applying spectral characteristics of the reference gradation value for the prediction target color to a corresponding second relational equation; andan interpolation step of obtaining prediction values of spectral characteristics of a gradation value not included in the union among two or more of the prediction target gradation values by performing spline interpolation based on the prediction values obtained in the second relational equation using step.9. The spectral characteristics prediction method according to claim 8, further comprising a third prediction step, between the union creation step and the second relational equation calculation step, of obtaining prediction values of spectral characteristics, which are spectral characteristics of the process target gradation value for the reference color, of a gradation value for which spectral characteristics are not obtained regarding the reference color by performing spline interpolation based on known spectral characteristics for the reference color.10. The spectral characteristics prediction method according to claim 1, wherein the difference value obtained in the difference value calculation step is a square error of the prediction values obtained in the first prediction step and the actual measurement values of the spectral characteristics of the characteristics-acquired gradation value for the prediction target color.11. The spectral characteristics prediction method according to claim 1, wherein the difference value obtained in the difference value calculation step is a color difference based on the prediction values obtained in the first prediction step and the actual measurement values of the spectral characteristics of the characteristics-acquired gradation value for the prediction target color.12. The spectral characteristics prediction method according to claim 1, wherein the spectral characteristics are any of spectral reflectances, spectral absorption factors, and spectral absorption coefficients.13. A non-transitory computer-readable recording medium recording a spectral characteristics prediction program of predicting spectral characteristics obtained by applying ink on a base material, whereinthe spectral characteristics prediction program causes a computer to execute:a prediction target color setting step of setting, to a prediction target color, an ink color for which spectral characteristics of a maximum gradation value, spectral characteristics of a minimum gradation value, and spectral characteristics of at least one intermediate gradation value are obtained;a first relational equation calculation step of obtaining a first relational equation that, with an intermediate gradation value for which spectral characteristics are obtained regarding the prediction target color being taken as a characteristics-acquired gradation value, represents a relationship between spectral characteristics of the maximum gradation value and spectral characteristics of the characteristics-acquired gradation value for each of a plurality of sample colors which are a plurality of ink colors for each of which spectral characteristics of the maximum gradation value, spectral characteristics of the minimum gradation value, and spectral characteristics of at least one intermediate gradation value are obtained;a first prediction step of, for each of the plurality of sample colors, obtaining prediction values of spectral characteristics of the characteristics-acquired gradation value for the prediction target color by applying the spectral characteristics of the maximum gradation value for the prediction target color to a corresponding first relational equation;a difference value calculation step of, for each of the plurality of sample colors, obtaining a difference value between the prediction values obtained in the first prediction step and actual measurement values of spectral characteristics of the characteristics-acquired gradation value for the prediction target color;a reference color selection step of selecting, as a reference color, a sample color for which a minimum difference value is obtained in the difference value calculation step among the plurality of sample colors;a second relational equation calculation step of, with the maximum gradation value or the characteristic-acquired gradation value being taken as a reference gradation value, and with a gradation value for which spectral characteristics are obtained regarding the reference color or the characteristics-acquired gradation value being taken as a process target gradation value, obtaining a second relational equation that represents a relationship between spectral characteristics of the reference gradation value and spectral characteristics of the process target gradation value for the reference color; anda second prediction of, using the second relational equation, obtaining prediction values of spectral characteristics of a prediction target gradation value for the prediction target color.